Development and Validation of a Computational Model for Androgen Receptor Activity.

Testing thousands of chemicals to identify potential androgen receptor (AR) agonists or antagonists would cost millions of dollars and take decades to complete using current validated methods. High-throughput in vitro screening (HTS) and computational toxicology approaches can more rapidly and inexpensively identify potential androgen-active chemicals. We integrated 11 HTS ToxCast/Tox21 in vitro assays into a computational network model to distinguish true AR pathway activity from technology-specific assay interference. The in vitro HTS assays probed perturbations of the AR pathway at multiple points (receptor binding, coregulator recruitment, gene transcription, and protein production) and multiple cell types. Confirmatory in vitro antagonist assay data and cytotoxicity information were used as additional flags for potential nonspecific activity. Validating such alternative testing strategies requires high-quality reference data. We compiled 158 putative androgen-active and -inactive chemicals from a combination of international test method validation efforts and semiautomated systematic literature reviews. Detailed in vitro assay information and results were compiled into a single database using a standardized ontology. Reference chemical concentrations that activated or inhibited AR pathway activity were identified to establish a range of potencies with reproducible reference chemical results. Comparison with existing Tier 1 AR binding data from the U.S. EPA Endocrine Disruptor Screening Program revealed that the model identified binders at relevant test concentrations (<100 µM) and was more sensitive to antagonist activity. The AR pathway model based on the ToxCast/Tox21 assays had balanced accuracies of 95.2% for agonist (n = 29) and 97.5% for antagonist (n = 28) reference chemicals. Out of 1855 chemicals screened in the AR pathway model, 220 chemicals demonstrated AR agonist or antagonist activity and an additional 174 chemicals were predicted to have potential weak AR pathway activity.

On selecting a minimal set of in vitro assays to reliably determine estrogen agonist activity.

The US EPA is charged with screening chemicals for their ability to be endocrine disruptors through interaction with the estrogen, androgen and thyroid axes. The agency is exploring the use of high-throughput in vitro assays to use in the Endocrine Disruptor Screening Program (EDSP), potentially as replacements for lower-throughput in vitro and in vivo tests. The first replacement is an integrated computational and experimental model for estrogen receptor (ER) activity, to be used as an alternative to the EDSP Tier 1 in vitro ER binding and transactivation assays and the in vivo uterotrophic bioassay. The ER agonist model uses a set of 16 in vitro assays that incorporate multiple technologies and cell lines and probe multiple points in the ER pathway. Here, we demonstrate that subsets of assays with as few as 4 assays can predict the activity of all 1811 chemicals tested with accuracy equivalent to that of the full 16-assay model. The prediction accuracy against reference chemicals is higher than that of the full chemical set, partly because the larger set contains many chemicals that can cause a variety of types of assay interference There are multiple accurate assay subsets, allowing flexibility in the construction of a multiplexed assay battery. We also discuss the issue of challenging chemicals, i.e. those that can give false positive results in certain assays, and could hence be more problematic when only a few assays are used.

Early Drug-Induced Liver Injury Risk Screening: "Free," as Good as It Gets.

For all the promise of and need for clinical drug-induced liver injury (DILI) risk screening systems, demonstrating the predictive value of these systems versus readily available physicochemical properties and inherent dosing information has not been thoroughly evaluated. Therefore, we utilized a systematic approach to evaluate the predictive value of in vitro safety assays including bile salt export pump transporter inhibition and cytotoxicity in HepG2 and transformed human liver epithelial along with physicochemical properties. We also evaluated the predictive value of in vitro ADME assays including hepatic partition coefficient (Kp) and its unbound counterpart because they provide insight on hepatic accumulation potential. The datasets comprised of 569 marketed drugs with FDA DILIrank annotation (most vs less/none), dose and physicochemical information, 384 drugs with Kp and plasma protein binding data, and 279 drugs with safety assay data. For each dataset and combination of input parameters, we developed random forest machine learning models and measured model performance using the receiver operator characteristic area under the curve (ROC AUC). The median ROC AUC across the various data and parameters sets ranged from 0.67 to 0.77 with little evidence of additive predictivity when including safety or ADME assay data. Subsequent machine learning models consistently demonstrated daily dose, fraction sp3 or ionization, and cLogP/D inputs produced the best, simplest model for predicting clinical DILI risk with an ROC AUC of 0.75. This systematic framework should be used for future assay predictive value assessments and highlights the need for continued improvements to clinical DILI risk annotation.

Use of high throughput ion channel profiling and statistical modeling to predict off-target arrhythmia risk - One pharma's experience and perspective.

INTRODUCTION: The use of high throughput patch clamp profiling to determine mixed ion channel-mediated arrhythmia risk was assessed using profiling data generated using proprietary internal and clinical reference compounds. We define the reproducibility of the platform and highlight inherent platform issues. The data generated was used to develop predictive models for cardiac arrhythmia risk, specifically Torsades de Pointes (TdP). METHODS: A retrospective analysis was performed using profiling data generated over a 3-year period, including patch clamp data from hERG, Cav1.2, and Nav1.5 (peak/late), together with hERG binding. RESULTS: Assay reproducibility was robust over the 3-year period examined. High throughput hERG patch IC50 values correlated well with GLP-hERG data (Pearson = 0.87). A disconnect between hERG binding and patch was observed for ∼10% compounds and trended with passive cellular permeability. hERG and Cav1.2 potency did not correlate for proprietary compounds, with more potent hERG compounds showing selectivity versus Cav1.2. For clinical compounds where hERG and Cav1.2 activity was more balanced, an analysis of TdP risk versus hERG/Cav1.2 ratio demonstrated low TdP probability when the hERG/Cav1.2 potency ratios were < 1. Modeling of clinical compound data revealed a lack of impact of the Nav1.5 (late) current in predicting TdP. Moreover, models using hERG binding data (ROC AUC = 0.876) showed an improved ability to predict TdP risk versus hERG patch clamp (ROC AUC = 0.787). DISCUSSION: The data highlight the value of high throughput patch clamp data in the prediction of TdP risk, as well as some potential limitations with this approach.

Reengineering rate-limiting, millisecond enzyme motions by introduction of an unnatural amino acid.

Rate-limiting millisecond motions in wild-type (WT) Ribonuclease A (RNase A) are modulated by histidine 48. Here, we incorporate an unnatural amino acid, thia-methylimidazole, at this site (H48C-4MI) to investigate the effects of a single residue on protein motions over multiple timescales and on enzyme catalytic turnover. Molecular dynamics simulations reveal that H48C-4MI retains some crucial WT-like hydrogen bonding interactions but the extent of protein-wide correlated motions in the nanosecond regime is decreased relative to WT. NMR Carr-Purcell-Meiboom-Gill relaxation dispersion experiments demonstrate that millisecond conformational motions in H48C-4MI are present over a similar pH range compared to WT. Furthermore, incorporation of this nonnatural amino acid allows retention of WT-like catalytic activity over the full pH range. These studies demonstrate that the complexity of the protein energy landscape during the catalytic cycle can be maintained using unnatural amino acids, which may prove useful in enzyme design efforts.

Characterization of a high throughput human stem cell cardiomyocyte assay to predict drug-induced changes in clinical electrocardiogram parameters.

Human induced pluripotent stem cell derived cardiomyocytes (hIPSC-CM's) play an increasingly important role in the safety profiling of candidate drugs. For such models to have utility a clear understanding of clinical translation is required. In the present study we examined the ability of our hIPSC-CM model to predict the clinically observed effects of a diverse set of compounds on several electrocardiogram endpoints, including changes in QT and QRS intervals. To achieve this, compounds were profiled in a novel high throughput voltage-sensitive dye platform. Measurements were taken acutely (30 min) and chronically (24 h) to ensure that responses from compounds with slow onset kinetics or that affected surface ion channel expression would be captured. In addition, to avoid issues associated with changes in free drug levels due to protein binding, assays were run in serum free conditions. Changes in hIPSC-CM threshold APD90 values correlated with compound plasma exposures that produced a +10 ms change in clinical QTc (Pearson r2 = 0.80). In addition, randomForest modeling showed high predictivity in defining TdP risk (AUROC value = 0.938). Risk associated with QRS prolongation correlated with an increase in action potential rise-time (AUROC value = 0.982). The in-depth understanding of the clinical translatability of our hIPSC-CM model positions this assay to play a key role in defining cardiac risk early in drug development. Moreover, the ability to perform longer term studies enables the detection of compounds that may not be highlighted by more acute assay formats, such as inhibitors of hERG trafficking.

The flexibility of a distant loop modulates active site motion and product release in ribonuclease A.

The role of the flexible loop 1 in protein conformational motion and in the dissociation of enzymatic product from ribonuclease A (RNase A) was investigated by creation of a chimeric enzyme in which a 6-residue loop 1 from the RNase A homologue, eosinophil cationic protein (ECP), replaced the 12-residue loop 1 in RNase A. The chimera (RNase A(ECP)) experiences only local perturbations in NMR backbone chemical shifts compared to WT RNase A. Many of the flexible residues that were previously identified in WT as involved in an important conformational change now experience no NMR-detected millisecond motions in the chimera. Likewise, binding of the product analogue, 3'-CMP, to RNase A(ECP) results in only minor chemical shift changes in the enzyme similar to what is observed for the H48A mutant of RNase A and in contrast to WT enzyme. For both RNase A(ECP) and H48A there is a 10-fold decrease in the product release rate constant, k(off), compared to WT, in agreement with previous studies indicating the importance of flexibility in RNase A in the overall rate-limiting product release step. Together, these NMR and biochemical experiments provide additional insight into the mechanism of millisecond motions in the RNase A catalytic cycle.

Characterizing complex dynamics in the transactivation response element apical loop and motional correlations with the bulge by NMR, molecular dynamics, and mutagenesis.

The HIV-1 transactivation response element (TAR) RNA binds a variety of proteins and is a target for developing anti-HIV therapies. TAR has two primary binding sites: a UCU bulge and a CUGGGA apical loop. We used NMR residual dipolar couplings, carbon spin relaxation (R(1) and R(2)), and relaxation dispersion (R(1rho)) in conjunction with molecular dynamics and mutagenesis to characterize the dynamics of the TAR apical loop and investigate previously proposed long-range interactions with the distant bulge. Replacement of the wild-type apical loop with a UUCG loop did not significantly affect the structural dynamics at the bulge, indicating that the apical loop and the bulge act largely as independent dynamical recognition centers. The apical loop undergoes complex dynamics at multiple timescales that are likely important for adaptive recognition: U31 and G33 undergo limited motions, G32 is highly flexible at picosecond-nanosecond timescales, and G34 and C30 form a dynamic Watson-Crick basepair in which G34 and A35 undergo a slow (approximately 30 mus) likely concerted looping in and out motion, with A35 also undergoing large amplitude motions at picosecond-nanosecond timescales. Our study highlights the power of combining NMR, molecular dynamics, and mutagenesis in characterizing RNA dynamics.

Uncertainty quantification in ToxCast high throughput screening.

High throughput screening (HTS) projects like the U.S. Environmental Protection Agency's ToxCast program are required to address the large and rapidly increasing number of chemicals for which we have little to no toxicity measurements. Concentration-response parameters such as potency and efficacy are extracted from HTS data using nonlinear regression, and models and analyses built from these parameters are used to predict in vivo and in vitro toxicity of thousands of chemicals. How these predictions are impacted by uncertainties that stem from parameter estimation and propagated through the models and analyses has not been well explored. While data size and complexity makes uncertainty quantification computationally expensive for HTS datasets, continued advancements in computational resources have allowed these computational challenges to be met. This study uses nonparametric bootstrap resampling to calculate uncertainties in concentration-response parameters from a variety of HTS assays. Using the ToxCast estrogen receptor model for bioactivity as a case study, we highlight how these uncertainties can be propagated through models to quantify the uncertainty in model outputs. Uncertainty quantification in model outputs is used to identify potential false positives and false negatives and to determine the distribution of model values around semi-arbitrary activity cutoffs, increasing confidence in model predictions. At the individual chemical-assay level, curves with high variability are flagged for manual inspection or retesting, focusing subject-matter-expert time on results that need further input. This work improves the confidence of predictions made using HTS data, increasing the ability to use this data in risk assessment.

Tiered High-Throughput Screening Approach to Identify Thyroperoxidase Inhibitors Within the ToxCast Phase I and II Chemical Libraries.

High-throughput screening for potential thyroid-disrupting chemicals requires a system of assays to capture multiple molecular-initiating events (MIEs) that converge on perturbed thyroid hormone (TH) homeostasis. Screening for MIEs specific to TH-disrupting pathways is limited in the U.S. Environmental Protection Agency ToxCast screening assay portfolio. To fill 1 critical screening gap, the Amplex UltraRed-thyroperoxidase (AUR-TPO) assay was developed to identify chemicals that inhibit TPO, as decreased TPO activity reduces TH synthesis. The ToxCast phase I and II chemical libraries, comprised of 1074 unique chemicals, were initially screened using a single, high concentration to identify potential TPO inhibitors. Chemicals positive in the single-concentration screen were retested in concentration-response. Due to high false-positive rates typically observed with loss-of-signal assays such as AUR-TPO, we also employed 2 additional assays in parallel to identify possible sources of nonspecific assay signal loss, enabling stratification of roughly 300 putative TPO inhibitors based upon selective AUR-TPO activity. A cell-free luciferase inhibition assay was used to identify nonspecific enzyme inhibition among the putative TPO inhibitors, and a cytotoxicity assay using a human cell line was used to estimate the cellular tolerance limit. Additionally, the TPO inhibition activities of 150 chemicals were compared between the AUR-TPO and an orthogonal peroxidase oxidation assay using guaiacol as a substrate to confirm the activity profiles of putative TPO inhibitors. This effort represents the most extensive TPO inhibition screening campaign to date and illustrates a tiered screening approach that focuses resources, maximizes assay throughput, and reduces animal use.

High-Resolution NMR Studies of Human Tissue Factor.

In normal hemostasis, the blood clotting cascade is initiated when factor VIIa (fVIIa, other clotting factors are named similarly) binds to the integral membrane protein, human tissue factor (TF). The TF/fVIIa complex in turn activates fX and fIX, eventually concluding with clot formation. Several X-ray crystal structures of the soluble extracellular domain of TF (sTF) exist; however, these structures are missing electron density in functionally relevant regions of the protein. In this context, NMR can provide complementary structural information as well as dynamic insights into enzyme activity. The resolution and sensitivity for NMR studies are greatly enhanced by the ability to prepare multiple milligrams of protein with various isotopic labeling patterns. Here, we demonstrate high-yield production of several isotopically labeled forms of recombinant sTF, allowing for high-resolution NMR studies both in the solid and solution state. We also report solution NMR spectra at sub-mM concentrations of sTF, ensuring the presence of dispersed monomer, as well as the first solid-state NMR spectra of sTF. Our improved sample preparation and precipitation conditions have enabled the acquisition of multidimensional NMR data sets for TF chemical shift assignment and provide a benchmark for TF structure elucidation.

Experimental Protein Structure Verification by Scoring with a Single, Unassigned NMR Spectrum.

Standard methods for de novo protein structure determination by nuclear magnetic resonance (NMR) require time-consuming data collection and interpretation efforts. Here we present a qualitatively distinct and novel approach, called Comparative, Objective Measurement of Protein Architectures by Scoring Shifts (COMPASS), which identifies the best structures from a set of structural models by numerical comparison with a single, unassigned 2D (13)C-(13)C NMR spectrum containing backbone and side-chain aliphatic signals. COMPASS does not require resonance assignments. It is particularly well suited for interpretation of magic-angle spinning solid-state NMR spectra, but also applicable to solution NMR spectra. We demonstrate COMPASS with experimental data from four proteins--GB1, ubiquitin, DsbA, and the extracellular domain of human tissue factor--and with reconstructed spectra from 11 additional proteins. For all these proteins, with molecular mass up to 25 kDa, COMPASS distinguished the correct fold, most often within 1.5 Å root-mean-square deviation of the reference structure.

Characterization of enzyme motions by solution NMR relaxation dispersion.

In many enzymes, conformational changes that occur along the reaction coordinate can pose a bottleneck to the rate of conversion of substrates to products. Characterization of these rate-limiting protein motions is essential for obtaining a full understanding of enzyme-catalyzed reactions. Solution NMR experiments such as the Carr-Purcell-Meiboom-Gill (CPMG) spin-echo or off-resonance R 1rho pulse sequences enable quantitation of protein motions in the time range of microseconds to milliseconds. These experiments allow characterization of the conformational exchange rate constant, k ex, the equilibrium populations of the relevant conformations, and the chemical shift differences (Deltaomega) between the conformations. The CPMG experiments were applied to the backbone N-H positions of ribonuclease A (RNase A). To probe the role of dynamic processes in the catalytic cycle of RNase A, stable mimics of the apo enzyme (E), enzyme-substrate (ES) complex, and enzyme-product (EP) complex were formed. The results indicate that the ligand has relatively little influence on the kinetics of motion, which occurs at 1700 s (-1) and is the same as both k cat, and the product dissociation rate constant. Instead, the effect of ligand is to stabilize one of the pre-existing conformations. Thus, these NMR experiments indicate that the conformational change in RNase A is ligand-stabilized and does not appear to be ligand-induced. Further evidence for the coupling of motion and enzyme function comes from the similar solvent deuterium kinetic isotope effect on k ex derived from the NMR measurements and k cat from enzyme kinetic studies. This isotope effect of approximately 2 depends linearly on solvent deuterium content suggesting the involvement of a single proton in RNase A motion and function. Moreover, mutation of His48 to alanine eliminates motion in RNase A and decreases the catalytic turnover rate indicating the involvement of His48, which is far from the active site, in coupling motion and function. For the enzyme triosephosphate isomerase (TIM), the opening and closing motion of a highly conserved active site loop (loop 6) has been implicated in many studies to play an important role in the catalytic cycle of the enzyme. Off-resonance R 1rho experiments were performed on TIM, and results were obtained for amino acid residues in the N-terminal (Val167), and C-terminal (Lys174, Thr177) portions of loop 6. The results indicate that all three loop residues move between the open and closed conformation at about 10,000 s (-1), which is the same as the catalytic rate constant. The O (eta) atom of Tyr208 provides a hydrogen bond to stabilize the closed form of loop 6 by interacting with the amide nitrogen of Ala176; these atoms are outside of hydrogen bonding distance in the open form of the enzyme. Mutation of Tyr208 to phenylalanine results in significant loss of catalytic activity but does not appear to alter the kex value of the N-terminal part of loop 6. Instead, removal of this hydrogen bond appears to result in an increase in the equilibrium population of the open conformer of loop 6, thereby resulting in a loss of activity through a shift in the conformational equilibrium of loop 6. Solution NMR relaxation dispersion experiments are powerful experimental tools that can elucidate protein motions with atomic resolution and can provide insight into the role of these motions in biological function.

Hydrogen exchange of monomeric alpha-synuclein shows unfolded structure persists at physiological temperature and is independent of molecular crowding in Escherichia coli.

Amide proton NMR signals from the N-terminal domain of monomeric alpha-synuclein (alphaS) are lost when the sample temperature is raised from 10 degrees C to 35 degrees C at pH 7.4. Although the temperature-induced effects have been attributed to conformational exchange caused by an increase in alpha-helix structure, we show that the loss of signals is due to fast amide proton exchange. At low ionic strength, hydrogen exchange rates are faster for the N-terminal segment of alphaS than for the acidic C-terminal domain. When the salt concentration is raised to 300 mM, exchange rates increase throughout the protein and become similar for the N- and C-terminal domains. This indicates that the enhanced protection of amide protons from the C-terminal domain at low salt is electrostatic in nature. Calpha chemical shift data point to <10% residual alpha-helix structure at 10 degrees C and 35 degrees C. Conformational exchange contributions to R2 are negligible at both temperatures. In contrast to the situation in vitro, the majority of amide protons are observed at 37 degrees C in 1H-15N HSQC spectra of alphaS encapsulated within living Escherichia coli cells. Our finding that temperature effects on alphaS NMR spectra can be explained by hydrogen exchange obviates the need to invoke special cellular factors. The retention of signals is likely due to slowed hydrogen exchange caused by the lowered intracellular pH of high-density E. coli cultures. Taken together, our results emphasize that alphaS remains predominantly unfolded at physiological temperature and pH-an important conclusion for mechanistic models of the association of alphaS with membranes and fibrils.

The mechanism of rate-limiting motions in enzyme function.

The ability to use conformational flexibility is a hallmark of enzyme function. Here we show that protein motions and catalytic activity in a RNase are coupled and display identical solvent isotope effects. Solution NMR relaxation experiments identify a cluster of residues, some distant from the active site, that are integral to this motion. These studies implicate a single residue, histidine-48, as the key modulator in coupling protein motion with enzyme function. Mutation of H48 to alanine results in loss of protein motion in the isotope-sensitive region of the enzyme. In addition, k(cat) decreases for this mutant and the kinetic solvent isotope effect on k(cat), which was 2.0 in WT, is near unity in H48A. Despite being located 18 A from the enzyme active site, H48 is essential in coordinating the motions involved in the rate-limiting enzymatic step. These studies have identified, of approximately 160 potential exchangeable protons, a single site that is integral in the rate-limiting step in RNase A enzyme function.

Resolving the motional modes that code for RNA adaptation.

Using a domain elongation strategy, we decoupled internal motions in RNA from overall rotational diffusion. This allowed us to site-specifically resolve a manifold of motional modes in two regulatory RNAs from HIV-1 with the use of nuclear magnetic resonance spin relaxation methods. Base and sugar librations vary on a picosecond time scale and occur within helical domains that move collectively at diffusion-limited nanosecond time scales. Pivot points are short, functionally important, and highly mobile internal loops. These spontaneous changes in RNA conformation correlate quantitatively with those that follow adaptive recognition of diverse targets. Thus, ligands may stabilize existing RNA conformations rather than inducing new ones.